20^th November 2001

National Physical Laboratory

 

Participants:		Lois Armstrong	Harwell Scientifics
		Roger Benzing	Imperial College
		Sylvie Chalamet	Harwell Instruments
		Peter Danyluk	AWE Aldermaston
		Ashley Davies	AWE Aldermaston
		Julian Dean	NPL
		Tony Dell	VLA
		Michael Froggatt	BNFL
		Arvic Harms	NPL
		Trevor Hatt	Perkin Elmer
		Ed Holden	Perkin Elmer
		Elizabeth Howie	British Energy, Hunterston
		Phil Hurst	British Energy, Heysham
		Simon Jerome	NPL
		Satwant Johal	NPL
		Richard Lukey	VLA
		Neil O'Brien	DSTL
		Desmond MacMahon	HMS Sultan
		Fiona Miller	UKAEA Dounreay
		Susan Parry	Imperial College
		Dan Parvin	BNFL Instruments
		Andy Pearce	NPL
		Martin Phillips	BNFL Berkeley
		Alison Raynor	AEA Technology QSA
		Martin Rushby	AWE Aldermaston
		Ian Sinclair	Harwell Instruments
		Bob Sharrock	Urenco
		Matthew Simpson	DSTL
		Harald Struwe	BNFL Instruments
		Robin Treadwell	AWE Aldermaston
		Tim Twomey	Perkin Elmer
		Stephanie Vallet	Harwell Instruments
		Karen Verrall	BNFL Berkley
		Tony Ware	Environmental Energy Consultancy
		Mike Woods	NPL (Chairman)
		Simon Woods	NPL (Secretary)
		Mike Youngman	NRPB

1. Chairman’s Opening Remarks

Mike Woods (Head of Science, Radioactivity Metrology Group, NPL) introduced himself as the new Chairman of the GSUF and welcomed the participants. He reviewed the GSUF activities over the preceding 3 years and invited the participants to provide feedback later in the meeting (see section 8).

2. Minutes of the last meeting

These were agreed as an accurate record.

3. Actions arising from the last meeting

Action 3.1:  NPL (Simon Woods) to amend and distribute questionnaires […on comparisons and low energy measurements].

Action 3.2: NPL (Simon Woods) to devise and distribute a suitable questionnaire […on measurement facilities].

Both actions have been completed, though the returns so far had been disappointing –

Comparisons – 3 returns
Low energy measurements – 4 returns
Facilities – 3 returns (2 laboratories)

Users were urged to return the questionnaires.

4. User discussions

4.1 Results of user surveys

Simon Woods gave a brief description of the received information to date.

4.1.1 Comparisons

User requirements are predominantly, though not exclusively in the environmental and bioassay fields, for at least 16 nuclides and chains over an activity range from less than 1 Bq/g up to 100 Bq/g, mainly in solid matrices. UKAS requirements mean that comparisons are desired annually.

4.2.2 Low energy measurements

Again the greatest need is in the environmental area over the full energy range. HPGe detectors seem to be exclusively used, normally calibrated with a mixed nuclide solution, with a wide range of measurement geometries. Most users do not model efficiency curves nor correct for attenuation in the sample.

4.2.3

Due to the sparse response so far, a web page will not yet be generated.

4.2 QA

How were manufacturers dealing with the QA of their software systems? There is a need to go further with their documentation. One route forward would be to generate more test spectra. Tim Twomey noted that there is a problem with test spectra if they are synthetic. If they are, then they are not a fair test because they do not use nuclear data and hence the software cannot use their data libraries to assist deconvolution. Also, in spectra such as those from the IAEA there are problems, such as the FWHM does not vary with energy.

The best option would be to generate test spectra from real sources but this then raises the problem of setting boundaries such as the mix of nuclides and background levels. Tim Twomey commented that, from a manufacturer’s point of view, it would be useful if NPL were to produce characterised spectra. The manufacturer could then include these with their software and documentation as validation, though “acceptability” would need to be defined, as no software is perfect.

Mike Woods asked if there was any literature on this subject available. Susan Parry remembered a paper from the US 6 or 7 years ago looking at environmental samples. Tim Twomey confirmed that Colin Sanderson was the author. However the problem is that the paper only describes how the software was working 7 years ago. Spectra should be produced once and for all and then the manufacturers could use them to test against for each software release.

Mike Woods suggested that NPL could produce standards and dilute them to the required levels, though there could be possible problems with matrix effects. Simon Jerome concurred and suggested that the methodology needs researching. There may also be a problem with finance in adjusting the NPL programme.

Mike Woods asked the users how widespread this need for more extension software QA was. Mike Froggatt replied that it would become more of a problem as users are driven to lower limits of detection. Tim Twomey added that it would also become more of a problem when uncertainties are fully considered. Test spectra would be a good starting point.

Roger Benzing queried the UKAS requirements. Mike Woods replied that the user must demonstrate that the software has been validated and that the procedures used are fit for purpose. This will vary between laboratories as each have different requirements. Tim Twomey added that assessors will also ask what comparisons are available and are they suitable. Susan Parry confirmed that this was a particular problem in activation analysis, as they can’t get reference materials for all of their work. Simon Jerome agreed. Assessors must be pragmatic as to whether or not suitable comparisons are available. Tim Twomey reiterated that having a starting point would be useful.

Roger Benzing was interested in testing software functions such as decay correction and efficiency/energy calibrations i.e. do the basic functions work correctly. Mike Woods suggested that a small group should get together with NPL to address boundary conditions, feasibility, nature of a suite of spectra, cost and effort.

Action 4.1: NPL (Simon Woods) to arrange a meeting on QA/Test spectra in January 2002

Interested parties are:	Perkin Elmer (EG&G Ortec)
	Harwell Instruments (Canberra)
	Roger Benzing
	Richard Lukey
	Mike Youngman
	Desmond MacMahon
	Matt Simpson

Tim Twomey asked if it were possible to issue a gamma-ray spectrum with a standard. Simon Jerome replied yes but queried as to which geometry.

4.3 Validation of Software

Susan Parry surmised that software has to be validated in some way as lines of software and asked what documentation was available from the manufacturers. Mike Woods added was the software validated against known standards. Susan Parry also asked whether all the mechanisms for validation were in place and working for the testing of basic software. Mike Woods further asked the manufacturers if there was an international testing programme standard.

Tim Twomey replied no. He and Ian Sinclair said they tested against known spectra. Simon Jerome queried whether this internal process could be made available to users. Tim Twomey observed that the forum was back to the standard spectra approach. With such an approach a document could be put together. Susan Parry queried whether there was an IEEE standard for software. Nobody present knew.

Trevor Hatt suggested that validation statements could probably be supplied but not the details. Susan Parry suggested that this was what she required; a piece of paper she could hand to an auditor. Tim Twomey agreed that a certificate could be supplied.

Martin Rushby stated that surely full documentation is performed and audited under the manufacturers’ own QA systems and that this could be provided to the users. Tim Twomey concurred that manufacturer’s should seek accreditation under ISO 17025. Roger Benzing added that a statement from the manufacturer as to what level the software has been tested should be supplied on purchase. Martin Rushby said that again the forum was back to the concept of needing test spectra.

The forum was agreed that at the next GSUF there should be talks from interested parties on this topic.

Action 4.2: Manufacturers (Perkin Elmer/Harwell Instruments) to address the 5^th GSUF on the QA and validation of their software.

Action 4.3: NPL (Simon Woods) to arrange for someone from their Software Validation Group to give their views on software QA and validation at the 5^th GSUF.

Tim Twomey asked if this validation should be extended to a system as a whole. Roger Benzing replied that this was the assessor’s job, but they often ask directly about the software.

4.4 Attenuation corrections

Richard Lukey asked the forum about methods used to measure these corrections, which could then be input into the analysis program. Practical experimental procedures were required. Trevor Hatt noted that GammaVision will input data for various types of source matrix. Richard Lukey enquired as to whether there was any validation between experiment and GammaTool.

Susan Parry stated that it was easier to use experimental data rather than individual calculations. Tim Twomey responded that with this approach there were sometimes major problems identifying the constitution of the sample. Ian Sinclair noted that Canberra have measured US reference material. Trevor Hatt added that one also needs to characterise each detector.

4.5 Standard sources

Martin Phillips explained that users were dependent on standard sources and particularly their quality. He had taken 3 or 4 manufacturers point sources and measured them for comparability to the stated activity. In some instances he was struggling to get agreement even at the 3s level. He has also been experiencing problems with efficiency curves constructed from the same source.

Mike Woods asked if the manufacturers had been contacted. Martin Phillips replied that they haven’t been as the sources are fit for purpose. He added that all of the certificates stated that the sources were traceable to national standards. Mike Woods said that he would like to see the data.

Action 4.4: Martin Phillips to send the measurement data to NPL (Simon Woods)

Mike Woods continued that, at the primary level, the national laboratories are seeking to demonstrate equivalence. At the secondary level a system of Certified Measurement Capabilities (CMCs) is operating. He queried whether people are quoting uncertainties lower than the declared national measurement capability.

Robin Treadwell asked as to the consistency of half-lives used and the longevity of sources. Mike Woods replied that the user should check the physical integrity of each source and perform a programme of leak testing. They should also consider whether the decay period is significant compared with the half-life, as this will cause the half-life uncertainties to multiply up. UKAS recommend recertification of a source every four years, or every two years if it is usedregularly. Martin Philips noted that integrity was a particular problem for Am sources.

5. NPL mixed nuclide source

Simon Jerome made a presentation outlining problems experienced with the current mixed nuclide composition and a set of proposed changes. The transparencies from this talk are appended to these minutes.

Trevor Hatt expressed concern at the proposed change from using ¹³⁹Ce (165 keV) to using ¹⁴¹Ce (145 keV), a nuclide with a far shorter half-life (32.5 days). These energies are at a critical point of the efficiency curve and use of the shorter lived nuclide may hinder re-use of the solution for further calibrations. Robin Treadwell commented that the mixed nuclide calibration could be supplemented by single nuclide calibrations to overcome this half-life problem. Simon Jerome advised that single nuclide solutions were available off the shelf from NPL.

Simon Jerome explained that the addition of ²¹⁰Pb was in addition to ²⁴¹Am in order to extend the calibration at low energies. Some users present expressed concern about the addition of further alpha activity (via the decay of ²¹⁰Po in the decay chain) to the solution.

Mike Woods enquired how often the users change their sources and recalibrate. Tony Dell informed the forum that VLA perform a full calibration each month. Mike Woods suggested that it should be possible to calibrate annually and then use the solution as a check source each month utilising the longer lived gamma emitters to overcome the half-life problem. Desmond MacMahon concurred. Mike Woods added that the production of a mixed nuclide solution helps to keep costs down; the production of single nuclide solutions is more expensive overall.

The users at the forum did not disagree to change the frequency of the mixed nuclide solution issue to six monthly (from four monthly). However, Tim Twomey, as a manufacturer, pointed out that if they sold a Ge detector at the “wrong time of year”, the change to the shorter lived isotopes would mean that the user could have problems with accurate calibration. Alison Raynor pointed out that this could be overcome as AEAT-QSA issue their mixed nuclide standard four times a year.

Martin Rushby commented that he found a solid version of the mix, recently supplied by NPL, to be easier to use than the solution.

6. Report of the WG on measurement uncertainties

Mike Woods and Simon Woods presented the work of the WG to date. The reports on the status of the deliberations are attached to these minutes.

Tim Twomey commented that the uncertainty, which might arise from using peak fitting as opposed to integration of the ROI, could be overcome if the same method was used for both calibration and assay of the sample. Mike Woods mused as to whether this was done.

The forum felt that it was important to differentiate between an uncertainty and a mistake e.g. tampering with data. Susan Parry suggested that the list of uncertainty sources could be split into two along the lines of Type A and Type B uncertainties.

Karen Verrall queried how do you address operator effects. Mike Woods suggested that it was necessary to conduct tests covering variety of operators in order to assess the variability between them and the accuracy and variability of individuals.

Desmond MacMahon raised the issue of the propagation of uncertainties and the treatment of correlations. Mike Woods replied that the best method of dealing with correlations is to avoid them; partial correlations are a very grey area. However, if this is not possible, then bounds can be set by using 100% positive and negative correlation.

The forum was happy with the way the WG was going and thought that a final publication of a GPG in this area would be a good idea as even though Ge detectors have been in existence since the mid sixties, there is still no simple users guide. Martin Phillips commented that it might be better to consider the topic of sampling in a separate GPG.

7. Discussions with manufacturers

Both Harwell Instruments (Stephanie Vallet) and Perkin Elmer (Tim Twomey) gave short updates as to the status of their companies and products.

7.1 Harwell Instruments

• Company update
• Apex Laboratory Productivity Suite
• Labsocs for Windows

Desmond MacMahon queried whether, for Labsocs, a comparison of source and sourceless calibrations had been made. Stephanie Vallet replied that it had and that there was full QA documentation. The sourceless calibrations were accurate to within 4 – 7%.

7.2 Perkin Elemer

• True coincidence summing corrections in GammaVision
• X-Cooler update

It was noted that the X-Cooler gave the same resolution as LN₂ above 500 keV, and no worse than 10% degradation below that energy.

8. Concluding discussions

At the beginning of the meeting Mike Woods gave an overview of previous topics addressed by the forum. These were now discussed in greater detail.

8.1 Terms of Reference

The forum did not see a need for an update.

8.2 Location of meetings

The forum was happy for the meetings to be continued to be held at NPL.

8.3 Nuclear data in The Radiochemical Manual

Desmond MacMahon suggested that the mistakes found in the data to date should be circulated.

Action 4.5: NPL (Simon Woods) to circulate mistakes to GSUF members and to enquire as to any updates to the list.

Lois Armstrong is in contact with Ian Adsley should the need arise to discuss further with him regarding an addendum. The question of ownership of the TRM has yet to be resolved as the work was performed by AEAT but funded by the DTI.

Action 4.6: NPL (Mike Woods) to contact DTI regarding ownership of the TRM.

Nuclear data should remain on the GSUF agenda.

8.4 Peak fitting

Roger Benzing explained that this had arisen when he was trying to fit a small ⁵⁴Mn peak which was lying on top of a large Compton continuum from ⁶⁰Co. Out of three packages, one consistently underestimated the area of the Mn peak, one consistently overestimated the area and one was reasonably ok. This was arising due to problems fitting the background. The problem has been resolved by simply adopting the latter analysis package.

Tim Twomey advised that he had found a similar problem in 1986 when analysing ^110mAg and ¹³⁴Cs post Chernobyl. EG&G had corrected the problem with an adaptive background approach.

This topic can now be considered closed.

8.5 QA/System validation

See sections 4.2 and 4.3 above. Item to remain on the agenda.

8.6 Field measurements

IIG didn’t want to liase with the GSUF. As it is a small interest group, NPL will discuss directly with those concerned.

Action 4.7: NPL (Simon Woods) to arrange discussion with Martin Phillips and Matt Simpson.

8.7 Uncertainties

An on-going topic – see section 6 above.

8.8 Manufacturers presentations

See action 4.2 above. The forum noted that so far it has concentrated on high resolution measurements. It would be interested in hearing updates on NaI and CdTe detector performance.

Action 4.8: Manufacturers to provide updates on detectors other than HPGe at the 5^th GSUF.

Action 4.9: All to advise Simon Woods if there are other topics that they feel should be addressed.

8.9 Questionnaires

The opportunity to return these questionnaires is open until the end of the year. After then, NPL will collate the results and produce a summary report.

Action 4.10: All to return questionnaires (if desired) to Simon Woods by 31/12/01.

Action 4.11: NPL to produce summary report of returned questionnaires.

8.10 Calibration

See section 5 above. Any further feedback would be welcome.

Action 4.12: All to provide feedback to Simon Jerome on suitable calibration sources.

9. Chairman’s closing remarks

Mike Woods thanked those present for attending and for their input to the forum.

 

Simon Woods

Secretary, GSUF